CircASH2L is reduced in HCC, and lower CircASH2L levels are associated with a worse prognosis in HCC patients
Two GEO datasets (GSE97332 and GSE78520) were analyzed to pinpoint the circular RNAs that are consistently disturbed in hepatocellular carcinoma (HCC). The analysis revealed that multiple circular RNAs exhibited decreased expression in HCC in comparison to precancerous tissues. In the GSE78520 dataset, 6 circRNAs were decreased and 55 circRNAs were increased in HCC (Fig.S1A). A total of 51 circRNAs were down-regulated and 98 circRNAs were up-regulated in the GSE97332 cohort (Fig.S1B). Analysis revealed 5 down-regulated and 16 up-regulated circRNAs in both cohorts (Fig.S1C). In HCC tissues, the circRNA hsa_circRNA_0006302 (circASH2L) was discovered to have a notable decrease in comparison to normal tissues (Fig.S1C). This circRNA, derived from exons 6, 7, 8, 9, and 10 of the ASH2L gene, has a length of 580 nt [CircBase ID: hsa_circ_0006302, termed circASH2L] and is identified as hsa_circ_0006302 in CircBase (Fig. 1A). The head-to-tail splicing of circASH2L was confirmed through Sanger sequencing (Fig. 1B). The sequence aligns with circBase database annotation. Convergent and divergent primers were developed to amplify ASH2L mRNA and circASH2L to exclude the possibility of head-to-tail splicing being caused by transsplicing or genomic rearrangements. Both cDNA and gDNA from HCC hepG2 cells were utilized as templates. The findings showed that circ0006302 could be amplified only in cDNA using divergent primers, and not in gDNA (Fig. 1C). qPCR examination of RNA level after RNase R treatment showed that circASH2L was resistant to degradation, while ASH2L mRNA transcripts were degraded (Fig. 1D and E). Moreover, treating HCC cells with actinomycin D, a transcription inhibitor, showed that the circASH2L transcript had a more stable half-life compared to ASH2L mRNA (Fig. 1F). The results collectively verified the attributes of circASH2L as a circRNA.
We assessed the level of circASH2L of 40 matched HCC samples and peri-tumor tissues (cohort 1) using qRT-PCR. RNA-seq data indicated a notable reduction in circASH2L levels in HCC samples compared to their corresponding non-cancerous tissues (Fig. 1G). CircASH2L level in HCC patients was further studied by determining its expression using in situ hybridization (ISH) on tissue microarrays (TMAs) including 122 pairs of HCC tissues (cohort 2). CircASH2L was markedly reduced in HCC specimens compared to the similar non-cancerous tissues, as seen in Fig. 1H and Fig.S1D. We next examined the connections between circASH2L expression and the clinicopathological features in 122 HCC patients, as shown in Table 1. The chi-square test indicated a significant association between decreased circASH2L levels in HCC tissues and various aggressive clinicopathologic indicators like microvascular invasion, advanced TNM stage, and BCLC stage. The survival plot indicated that individuals with hepatocellular carcinoma (HCC) who had lower levels of circASH2L expression experienced poorer overall survival rates (OS P = 0.007) and disease-free survival rates (DFS P = 0.017), as illustrated in Fig. 1.I and J). Our research indicates that circASH2L is reduced in HCC tissues and lower circASH2L levels in HCC are associated with a worse prognosis for patients.
Table 1
Correlation of the clinicopathological variables with CircASH2L expression in HCC
Clinicopathological | | CircASH2L expression | |
n | low | high | P Value |
Gender* | | | | | | |
Male | 102 | 54 | 48 | 1.000 |
Femanle | 20 | 11 | 9 |
Age* | | | | | | |
≤ 50 | 59 | 35 | 24 | 0.209 |
> 50 | 63 | 30 | 33 |
ALT(ng/mL)* | | | | | | |
≤ 75 | 113 | 62 | 51 | 0.294 |
> 75 | 9 | 3 | 6 |
AFP(ug/ul)* | | | | | | |
≤ 20 | 32 | 11 | 21 | 0.013 |
> 20 | 90 | 54 | 36 |
Cirrhosis* | | | | | | |
NO | 34 | 18 | 16 | 0.963 |
YES | 88 | 47 | 41 |
Tumor size* | | | | | | |
≤ 5 | 49 | 24 | 25 | 0.436 |
> 5 | 73 | 41 | 32 |
Tumor encapsulation* | | | | | | |
Complete | 66 | 27 | 39 | 0.003 |
None | 56 | 38 | 18 |
Tumor number* | | | | | | |
Solitary | 95 | 50 | 45 | 0.788 |
Multiple(≥ 2) | 27 | 15 | 12 |
TNM clinical stage | | | | | | |
Ⅰ-Ⅱ | 91 | 43 | 48 | 0.022 |
Ⅲ-Ⅳ | 31 | 22 | 9 |
Vascular invasion* | | | | | | |
NO | 97 | 48 | 49 | 0.098 |
YES | 25 | 17 | 8 |
Differentiation* | | | | | | |
Ⅰ-Ⅱ | 42 | 35 | 7 | 0.001 |
Ⅲ-Ⅳ | 80 | 30 | 50 |
BCLC stage* | | | | | | |
A | 80 | 37 | 43 | 0.032 |
B + C | 42 | 28 | 14 |
Abbreviations: TNM, tumor node metastasis; BCLC: Barcelona Clinic Liver Cancer. Bold text represents P-values with significant difference. |
CircASH2L hinders the growth, the migratory and invasive properties of HCC cells in vitro
Subsequently, the biological roles of circASH2L in HCC advancement were examined. We selected the HCC cell lines for silencing or overexpression of circASH2L based on the analysis of circASH2L expression using RT-qPCR in several human HCC cell lines. MHCC97H, HLF, and HCCLM3 cells showed reduced levels of endogenous circASH2L, while HepG2, Hep3B, Huh7, Sk-hep1, and PLC/PRF/5 cells displayed increased circASH2L expression (Fig.S2A). Thus, MHCC97H and HLF cells were chosen for upregulating circASH2L, whereas HepG2 and Huh7 cells were selected for downregulating circASH2L. Subsequently, we generated stable clones of MHCC97H and HLF cells with considerable overexpression of circASH2L (Fig. 2A). We successfully reduced circASH2L expression in HepG2 and Huh7 cells by employing small interfering RNAs (siRNAs) that target the back-splicing region, as shown in Fig. 2B. We used qPCR to demonstrate that our transfection did not impact additional ASH2L splicing variants. ASH2L mRNA expression remained unchanged after either overexpression or silencing of circASH2L, indicating the specificity of circASH2L manipulation (Fig.S2B, C).
A series of experiments were carried out to assess the influence of circASH2L on cellular proliferation. Elevated circASH2L levels were found to notably decrease the growth of MHCC97H and HLF cells, as demonstrated by CCK-8 assay and colony formation experiments (Fig. 2C and E, Fig.S2D and F), while decreasing circASH2L levels promoted cell growth (Fig. 2D and F, Fig.S2E and G). EdU experiments showed that overexpressing circASH2L significantly reduced the number of EdU-positive cells (Fig. 2G, I), whereas reducing circASH2L had the reverse effect (Fig. 2H, J). In HCC cells, the study of the cell cycle revealed that circASH2L expression impeded the transition from G1 to S phase (Fig. 2K). In contrast, reducing circASH2L levels promoted this progression. (Fig. 2L). Following this, the influence of circASH2L on the migratory and invasive properties of cancer cells was evaluated through transwell and wound healing tests. Upregulation of circASH2L significantly impeded the migratory and invasive properties of MHCC97H and HLF cells (Fig. 2M and O, Fig.S2H). Conversely, suppressing circASH2L notably enhanced these properties of HepG2 and Huh7 cells (Fig. 2N and P, Fig.S2I).
CircASH2L inhibits HCC growth and metastasis of HCC cells in vivo
We further conducted a study to examined the impact of circASH2L on tumor development and metastasis in living organisms. MHCC97H cells after transfection with circASH2L plasmid or a control vector, were implanted subcutaneously in BALB/c nude mice. Within 21 days of injection, the findings showed that overexpressing circASH2L led to a significant decrease in xenograft tumor sizes and weights (Fig. 3A-C). Silencing circASH2L increased tumor development of HepG2 cells (Fig.S3A). CircASH2L knockdown dramatically increased the tumor sizes and weights (Fig.S3B and C). IHC was conducted after collecting the subcutaneous tumor tissues. Ki-67 levels were considerably reduced in xenograft tumors overexpressing circASH2L (Fig. 3D) and raised in tumors where circASH2L was knocked down (Fig.S3D and E). Consistently, circASH2L-overexpressed MHCC97H cells and vector control cells were orthotopically transplanted into the livers of nude mice for the intrahepatic tumor implantation experiment. The tumors were reduced in size in the group overexpressing circASH2L compared to the group with vector control after growing in situ for 4 weeks in HCC MHCC97H xenografts (Fig. 3E-I).
We then examined whether circASH2L may impact metastasis in a live organism. Cells that overexpressed circASH2L and negative control cells were injected intravenously in the tail vein of nude mice to assess lung metastasis. In nude mice injected with circASH2L-overexpressed HLF cells, tumor volume, number of nodules, and lung metastasis incidence decreased significantly compared to mice injected with control cells (Fig. 3J and K). This was determined by analyzing bioluminescence intensity, number of lung nodules (Fig. 3M-O), and incidence of lung metastasis (Fig. 3L). Overall, our research suggests that circASH2L plays a role in inhibiting the growth and metastasis of HCC cells in vivo.
CircASH2L could act as a miR-525-3p sponge
FISH was performed on HCC cells to investigate circASH2L expression at the subcellular level. The superimposed images displayed a predominant cytoplasmic localization of circASH2L expression (Fig. 4A). Subsequent to cellular isolation, both nuclear and cytoplasmic RNA samples were subjected to analysis. The quantification of circRNA expression was conducted by qRT-PCR, revealing a cytoplasmic predominance of circASH2L expression (Fig. 4B). We investigated whether circASH2L, which is situated in the cytoplasm, may bind to miRNAs, considering the known role of circRNAs as miRNA sponges. We performed RNA immunoprecipitation (RIP) using an antibody targeting argonaute 2 (AGO2) in HepG2 and Huh7 cells. The findings indicated that circASH2L was notably concentrated by the AGO2 antibody (Fig. 4C, Fig.S4A), indicating that circASH2L could serve as a binding site for AGO2 and miRNAs. The Targetscan prediction engine on the CircInteractome website (https //circinteractome.nia.nih.gov/) was utilized to discover miRNAs that could potentially target circASH2L. We identified 27 candidate miRNAs that are predicted to bind to circASH2L.We conducted a circRNA pull-down test using biotin-labeled circASH2L probes to analyze 27 potential miRNAs in the complex for precise identification of interacting miRNAs (Fig. 4E). qRT-PCR analysis revealed a significant increase in miR-524-3p and miR-525-3p levels in the circASH2L probe group compared with the control (Fig. 4D and E), suggesting that these miRNAs are connected with circASH2L in HCC cells. To confirm this discovery, a luciferase assay was performed using HEK-293T cells co-transfected with circASH2L-wt reporter plasmid alongside miR-524-3p or miR-525-3p mimics. The results indicated that miR-525-3p imitations caused a notable reduction in luciferase activity in contrast to the control imitations NC, while miR-524-3p did not impact the luciferase activity (Fig. 4F). Consistent with the circRNA pull-down findings, miR-525-3p was identified as the highest enriched microRNA, suggesting its ability to bind circASH2L among the 27 microRNAs.
Next, we altered the specific locations targeted by miR-525-3p in the luciferase reporter. After transfection, a notable decrease in luciferase activity was observed in the wild-type (circASH2L-WT) HCC cell group following treatment with miR-525-3p mimics, whereas no impact was detected in circASH2L-mut group (Fig. 4G and H). This indicates a potential direct interaction between circASH2L and miR-525-3p. We conducted a pull-down test employing biotin-labeled miR-525-3p mimics and found a significant enrichment of circASH2L compared to the negative control (mimics NC) (Fig. 4I-K). Colocalization between circASH2L and miR-525-3p in HCC cells and xenograft tumor tissues was confirmed through a double FISH test (Fig. 4L and M). Nevertheless, circASH2L remained unchanged after the increase or decrease of miR-525-3p mimic expression in HCC MHHC97H and HepG2 cells, and likewise, miR-525-3p did not display notable variations after the enhancement or inhibition of circASH2L expression in HCC cells (Fig.S4B-E). This indicates that circASH2L and miR-525-3p may not degrade each other. Collectively, our studies indicate that circASH2L may act as a miR-525-3p sponge.
CircASH2L upregulates MTUS2 expression via sponging miR-525-3p
MiRNAs are widely recognized for their role in controlling the functions of target genes, ultimately influencing the behavior of cancer cells. We explored the downstream target of miR-525-3p to examine the pathogenic processes of circASH2L after discovering that miR-525-3p may be sponged by circASH2L. We used miRNA prediction techniques (miRDB, TargetScan, and miRwalk) to forecast the target mRNAs of miR-525-3p. It targeted nine candidates by overlapping with the results of miRNA target prediction, as illustrated in Fig. 5A. Following this, we elevated the levels of miR-525-3p mimics and examined the levels of their particular targets through qRT-PCR analysis (Fig.S5A, B). Genes suppressed by miR-525-3p include KCNQ5, MTUS2, AFAP1, TLX3, and RANBP10, among others. Following the overexpression or silencing of circASH2L, we subsequently monitored the levels of these targets. MTUS2 exhibited the most significant upregulation when circASH2L was overexpressed (Fig.S5C, D) and the most significant downregulation when circASH2L was shut down (Fig.S5E, F). Luciferase assays were employed to validate if miR-525-3p targeted MTUS2 mRNA. We created luciferase reporters with both wild type and mutant potential binding sites of MTUS2 transcripts (Fig. 5B). Luciferase reporter assays demonstrated a notable reduction in luciferase activity when miR-525-3p mimics were transfected, compared to the mimic NC, indicating the impact on the wild-type MTUS2 3′-UTR-wt (Fig. 5C). In contrast, the mutant luciferase reporter(MTUS2 3′-UTR -mut) was not affected (Fig. 5C). The results suggest a direct relationship between miR-525-3p and MTUS2 mRNA in HCC cells. An RNA-pulldown experiment was performed to confirm if circASH2L functions as a ceRNA in regulating MTUS2 expression by examining its influence on the binding of MTUS2 with miR-525-3p. The qRT-PCR findings indicated a significant decrease in the endogenous MTUS2 pull-down by the biotin-labeled miR-525-3p probe in cells transfected with circASH2L in comparison to those transfected with vector plasmids (Fig. 5D). This suggests that circASH2L may impede the interaction between MTUS2 and miR-525-3p. A luciferase assay demonstrated that altering the levels of circASH2L can lead to either an increase or decrease in luciferase activity of the MTUS2 wild type reporter (Fig. 5E). In addition, it was found that the introduction of circASH2L was able to recover the reporter activity of the normal MTUS2-3′-UTR -wt plasmid in HCC cells co-transfected with miR-525-3p mimics, but not in those with the mutated version (Fig. 5F). Western blot analysis indicated that increased miR-525-3p or decreased circASH2L led to a decrease in MTUS2 expression levels (Fig. 5G). Conversely, inhibiting miR-525-3p or increasing circASH2L expression led to a substantial rise in MTUS2 levels in HCC cells (Fig. 5H). The rise in MTUS2 caused by circASH2L overexpression was reversed by miR-525-3p mimics, as anticipated (Fig. 5I). These findings demonstrate that circASH2L functions as a decoy for miR-525-3p to control MTUS2 in HCC cells by relieving its suppression.
MiR-525-3p promotes HCC progression and reverses the tumor suppressor role of circASH2L in HCC cells
We conducted experiments to investigate the functional impact of miR-525-3p in order to gain a deeper understanding of its role. The CCK-8 experiments showed that upregulation of miR-525-3p led to enhanced proliferation of HCC cells, as seen in Fig. 6A-B. Transwell migration and matrigel invasion studies showed that miR-525-3p mimics significantly enhanced HCC cell migration and invasion (Fig. 6C-D). Furthermore, miR-525-3p were able to counteract the inhibitory effects of circASH2L on proliferation, migration, and invasion in MHCC97H cells (Fig. 6E-F). We altered the miR-525-3p-binding site on circASH2L and then transfected the modified circASH2L-miR-mut into MHCC97H cells to confirm that the cellular phenotype was induced by the interaction between circASH2L and miR-525-3p. Surprisingly, the results showed no significant distinction in the properties of cells to grow, move, and invade when transfected with circASH2L-miR-mut or control vector (Fig. 6G-H). These findings reinforce the idea that upregulation of miR-525-3p accelerates progression of HCC and hinders the tumor-inhibiting role of circASH2L in HCC cells.
CircASH2L suppresses HCC progression through the miR-525-3p-MTUS2 axis
Our hypothesis is that MTUS2 is involved in the tumor suppression process of the circASH2L-miR-525-3p axis, leading to the inhibition of malignant biological behaviors. Examination of the TCGA dataset showed a significant reduction in MTUS2 messenger RNA levels in HCC samples when compared to healthy tissues (Fig.S6A). Lower MTUS2 mRNA levels were linked to worse overall survival in liver cancer patients, as seen in the Kaplan-Meier plots (Fig.S6B). To delve deeper into the role of MTUS2, we investigated the functional impact of MTUS2 on HCC cells. Inhibition of MTUS2 was found to increase the ability of HepG2 and Huh7 cell lines to proliferate, migrate, and invade, as evidenced by Fig. 7A-D. The findings above identified MTUS2 as a tumor suppressor gene. We then designed rescue experiments. Overexpression of circASH2L in MHCC97H cells resulted in suppression of proliferation, migration, and invasion effects, which were reversed by knocking down MTUS2 (Fig. 7E-F). To confirm whether these findings could be replicated in a living organism, we introduced a xenograft hepatocellular carcinoma (HCC) model. The tumor suppression caused by circASH2L overexpression was successfully reversed by MTUS2 knockdown, as shown in Fig. 7G-I, consistent with the in vitro experiment findings. utilizing qRT-PCR method, we identified the presence of MTUS2 and circASH2L expression in the 40 HCC tissues. Pearson correlation analysis revealed a significant positive correlation between MTUS2 and circASH2L levels (Fig. 7J). The findings together show that circASH2L inhibits the development of HCC via the circASH2L / miR-525-3p / MTUS2 axis.
CircASH2L is modulated by m6A RNA methylation
The m6A alteration is crucial for posttranscriptional control and the formation of circRNAs[19]. We used RMBase 2.0 to identify probable m6A sites in circASH2L and identified 8 m6A sites using the algorithm (Fig.S7A). We used methylated RNA immunoprecipitation (MeRIP) tests to isolate known m6A-containing RNAs like RARA and circASH2L, which also showed significant m6A methylation (Fig. 8A), confirming the presence of m6A methylation in circASH2L. The abnormal expression of m6A writers (METTL3, METTL14, and WATP) is crucial for m6A modification and is linked to the aggressive development of HCC[20, 21]. We silenced m6A writers individually and found that only METTL3 knockdown, not METTL14 or WTAP knockdown, substantially influenced circASH2L expression, indicating METTL3 as the primary regulator of m6A modification in circASH2L regulation (Fig.S7B-D). METTL3 is the primary methyltransferase responsible for m6A alteration and contributes to the development of tumors[22]. RIP experiments confirmed the presence of circASH2L in complexes isolated with METTL3 antibody compared to those isolated with control IgG (Fig. 8B). To validate the interaction between METTL3 and circASH2L, we performed an RNA pull-down experiment as shown in Fig. 8C. M6A-specific immunoprecipitation assays (MeRIP) demonstrated a substantial drop in the m6A levels of circASH2L when METTL3 was silenced (Fig. 8D). Conversely, up-regulation of METTL3 of HCC cells resulted in a notable rise in circASH2L m6A levels (Fig. 8E). Thus, we indicated that METTL3 mediates the m6A modification of circASH2L in HCC. We then questioned if the m6A methyltransferase METTL3 may influence circASH2L function. When METTL3 was silenced, the half-life of circASH2L was extended compared to the control scramble (Fig. 8F), leading to a considerable increase in the expression of circASH2L (Fig. 8G). Enforcing the expression of METTL3 decreased the stability of circASH2L and lowered its expression levels in HCC cells (Fig. 8H and I). In 70 HCC patient samples, an immunohistochemistry (IHC) analysis showed a reverse relationship between circASH2L expression and METTL3 expression levels (8J and K). The findings showed that the METTL3-mediated m6A alteration may control circASH2L expression by aiding in its degradation.